Open eschava opened 7 years ago
It is not surprising me, I stopped working on it for the moment. I think it is not suffisant for production. Have you find a fix ?
No, that ATN tree walking is kind of miracle for me But I see that "lang" autocompletion uses another approach. Is there any article describing both ways?
Langs that are herited from AbstractLangEx
are old version.
With this article http://www.soft-gems.net/index.php/tools/47-universal-code-completion-using-antlr , i made langs herited from AbstractLang
that normally work better.
You can check next implementations of auto-completion libraries for antlr to find required code https://github.com/mike-lischke/antlr4-c3 (TypeScript) https://github.com/ftomassetti/antlr4-c3-kotlin (Kotlin)
Thanks for the links, I will check it ;)
Expression (1)+ has candidates [TAN, LN, Ident, Number, LOG, COS, ACOS, LPAREN, SIN, ASIN, ATAN]
but expression (1+1)+ has empty set of candidates
Another related thing: Expression (1 has candidates [DIV, TIMES, E, LT, POW, RPAREN, EQ, GT, PLUS, MINUS] (with RPAREN)
but expression (1+1 has only [DIV, TIMES, E, POW, PLUS, MINUS] (w/o RPAREN)